Cisco IT EMEAR Data Centre Day highlighted the technology path Cisco IT has taken toward a far Faster IT, and Digitization. And most of it is taking place in our Data Centre today.
Like most, every major business advancement comes from a mix of people, processes and technology. Each impacts the other. Where to begin?
Faster IT started for Cisco IT with server virtualization in the early 2000’s. Virtualization reduced the drain on data centre power and space (and server costs) as the demand for new applications grew. Today, Cisco IT still supports a range of virtualization hypervisors. The number of business applications running on virtual servers today is more than 95%. This level of virtualization is unusual (though not unique); but it enabled a dramatic shift to private cloud through automation.
In 2007, Cisco started migrating all their business applications, virtualized or not, to a new Cisco Nexus and UCS architecture. This move, along with the significant level of virtualization, reduced their total cost of ownership (TCO) of business applications within the data centre by more than 30%. Moreover, Cisco UCS enabled Service Profiles, the 30-odd pieces of blade server information that VM’s need to enable them to move to a new location inside a heterogeneous data centre floor. Setting up this Service Profile, and installing the VM and Operating System (OS) could be scripted, or automated. This automation of new VM’s started the first wave of private cloud inside Cisco. Developers could use resources as needed and then put them back “in the cloud” when done, and utilization improved.
Today Cisco IT’s cloud is based on Cisco Intelligent Automation for Cloud. Cisco IT developed Infrastructure as a Service (IaaS) and provided new server VM’s to employees in under 15 minutes, when previously it took 3 weeks to deliver the same virtual platform. This reduced operational costs tremendously; compute and storage TCO for virtual machines on the private cloud dropped by more than 30% once again.
But application developers don’t often develop directly onto a virtual server; they require libraries of middleware, and platforms (web portal or database or other platforms – there are hundreds available). So Cisco IT incorporated the delivery of complex platforms as part of this same automation/orchestration process within our private cloud, based on early containers from Red Hat OpenShift, again delivered in minutes. Among these more recent and more complex platforms are “Software Delivery as a Service” platforms. These platforms provide unique sets of automated code compilation, management, testing and deployment tools which enable our application developers to begin Continuous Delivery development of Java, Oracle, and other code within minutes. There are also “Hadoop as a Service” and “Splunk as a Service” platforms, which enable new business intelligence that are key to many digitization stories.
Agile development and the cloud-based continuous delivery has become the platform of choice for most new applications inside Cisco IT. Almost twice as many business services in Cisco are getting delivered with the same number of developers, almost 40% faster than before (and with improved quality and security). The “Mode 2” of bi-modal IT is getting faster, and for a short while the gap between those two modes, between slower IT infrastructure services and faster IT application-based services was getting wider.
But parts of the slower infrastructure services are about to move to the faster programmable side of bi-modal IT, starting with Cisco IT’s migration to Application Centric Infrastructure (ACI). With ACI the network services inside the data centre become centrally managed, programmatic, and orchestrated. Applications can be deployed with virtual cloud or physical servers, storage, switches, and networking services like firewalls, load balancing and QoS. There are long lists of other benefits – better resource utilization resulting in 25% lower capital costs, and an architecture that enables completely new security. But as the topic is Faster IT, I’ll just highlight two capabilities: applications are deployed using application-specific profiles automatically on the infrastructure for faster deployment; and end-to-end application monitoring enables faster network-based problem isolation and repair. This speed results in much lower operational costs (by over 45%).
There were hints at this EMEAR IT Data Centre Day of new plans to expand this network programmability beyond the walls of the data centre and into the wider area network as a whole. But not much was revealed, so that will have to wait for another Fast IT story.